I’ve been converting old German radio shows on casette tapes to CD’s (for use in a high school German course), and while most of the time I can just do a recording and then burn it with minimal editing, one show has been giving me some issues. It has what I can only describe as Charlie Brown voices (like the teacher, the Wah waah woh sound). I’ll post up a sample, but I’m not sure how (and if) this can be corrected.
My first attempt was just to cut out the lower frequencies through the equalizer. That made male voices sound much better, but didn’t do much for the female ones.
On my very cheap/tinny headphones, the male voices sound fine anyway, but on the surround sound system that the recordings are normally played on, I’m told it was completely unintelligible. I haven’t heard it first hand, but hopefully I will soon. I know there’s nothing dramatically wrong with the speaker system though.
Apparantly flac isn’t allowed for file attachments? What should I upload the sample as?
That’s the raw recording with no editing. It could just be that the computer playing the recording was set up to lift the bass or something, and that made everything unintelligible. Let me know what you think.
Oh, and let me know if you think the quality from my tape deck is lacking. It’s old, and one channel is actually burned out, if I remember correctly. I didn’t think there was an unacceptable noise level though.
There is quite a lot of background noise - hiss and people talking in the background - this combination makes it rather difficult to do much about the noise, but hopefully that will not be too much of a problem. The main issue is that the frequency bandwidth is rather limited with very little above about 3kHz. 3kHz is about the limit of old telephones, and is on the edge of being adequate for easily intelligible speech. It would have been nice to have frequencies up to at least 5kHz, but we will have to make the best of what we have.
Yes, the bass is rather too pronounced. If we look at the spectrum, we can see quite a lot of sound energy at around 30Hz. We can safely remove those frequencies as they have nothing to do with the voices. Also, the large peaks at around 150Hz and 300Hz add little to the intelligibility of the spoken word and can be safely reduced.
Where we are lacking is above 1.5 kHz, so we would want to boost those frequencies a bit.
As a side note, you may also notice a small spike at exactly 50Hz. This is caused by interference from mains electricity - so called “mains hum”. This shows that the recording was probably made in Europe (the mains frequency in the UK and mainland Europe is 50Hz, whereas in the US it is 60Hz).
I’ve not posted the Spectrum Window for the female voice, but it shows a pronounced peak at about 1kHz.
Here is the Equalizer effect window with settings to compensate for the above features:
Here is the file after Equalization. Not perfect, but I think you will find it is much clearer. http://www.sendspace.com/file/k78k7q
